Trading Systems in 'New Concepts In Technical Trading Systems' by J. Welles Wilder

[kaalilaatikko]

Originally Posted by dpaterso

One thing that is STILL concerning me (even MORe so now that kaalilaatikko and I have been thinking about this 'limit'): the 'limit' is NOT a number of points or pips it is a USD VALUE which AGAIN starts me wondering about whether or not the 60 point ASI is applicable or rather a 'correct value' ONLY for instruments or pairs the price of which is quoted in USD and this is what the original SI system was based on or designed around!!!

Now when we do not have explicit trading limits on pairs, we should use some virtual limits instead, assuming that fluctuations in price would not exceed these limits with very high probability. And if this would happen, you would just get an occasional value that is > 100, but the situation should restore to normal quite soon.

If I'm correct, this limit value is indeed not a pip value nor a value in a certain fixed currency (like USD), but a price difference in the quote currency of a pair.

What would you think about the following to calculate that limit: calculate the maximum for each plus/minus factor for a period that is representative enough to determine a virtual absolute maximum, e.g. half a year. Then take the maximum of those maximums and use that value as such as L.

You would need to check the validity of that L every now and then, as it might have risen or dropped.

J.

[dpaterso]

EUREKA!!!

'I've found it'!!! I HOPE!!!

I've attached a .zip file that contains two charts (both of EUR/USD daily). One chart shows the SI WITH the 'limit' as per 'the book' and MY SI with the 'limit' simply removed (accomplished by just leaving out the last portion of the equation i.e. the 'K/L' portion). The second chart shows the resulting ASI of both of the SI calculations mentioned above.

Now (and this is where I'd REALLY like some more opinions on the matter): I would imagine that JUST AS LONG AS ANY CHANGES do not affect the 'shape' of the 'plot' of both the SI and ASI then they are still 'valid'??? Would you agree with that???

Now in my attached examples what is interesting to note is this:

The 'shape' of the 'plot' of both the SI and the ASI are for the most part identical. The ONLY TIME when I can see a difference in the 'shape' of the 'plot' is when there has been an extreme move in price during the course of a day (and maybe over consecutive days). What this is telling me is that by simply removing the 'limit' from the equation my SI and ASI are no longer 'smoothed' (for want of a better word) on those days which, in MY mind, should be quite correct i.e. MY SI and ASI will now reflect the ENTIRE move in a day as opposed to only the allowable 'limit' move. DO YOU AGREE WITH ME???

Also (as I noted above): the 'shape' and the 'plot' of both the SI and the ASI are for the most part identical BUT they are EXACT at all HSP's and LSP's so the 'swing' of the SIS is 'preserved' and is intact!!! DO YOU AGREE WITH ME???

Even MORE interesting to note is the fact that by simply removing the 'limit' the SI appears to be 'correct' and 'standard arcoss the board' WITHOUT having to use QPF's or QPM's or PIPFACTORS or currency conversions or anything else!!! DO YOU AGREE WITH ME???

And BEST OF ALL: no matter WHAT instrument or pair you use MY SI and ASI on NOW the 60 point Trailing Index SAR 'appears' to 'appear' at a 'faesable' point i.e. it's no longer MILES AND MILES away on one instrument or pair and as close as makes no difference on another instrument or pair!!! DO YOU AGREE WITH ME???

(OK: well I'm going to modify the code and email all my 'clients' with THIS update AGAIN. Please take a look and see what I'm getting at).

[dpaterso]

NO NO!!!

kaalilaatikko:

Good idea BUT see my previous post!!!

[dpaterso]

Hokay!!!

Well: it's 'official' (again)!!!

I've just added the 'new' SIS i.e. with the 'limit' moves removed to a wide variety of instruments or pairs. If does not matter whether I'm looking at the Dow, GBP/JPY, Gold, or EUR/USD: the SIS Trailing Index SAR appears to be 'in touch and in tune' the with the ASI WITHOUT any 'manipulation' of the price formats etc. etc. etc. Also: the HSP's and LSP's are in exactly the same places as they were WITH the 'limit' being included so I'm confident that these are correct as well and that the 'operation' and / or 'design' of the SIS has not been compromised in any way whatsoever.

I think that THIS time: it's DONE!!!

Sorry for all the 'blips on the radar' BUT I reckon the SIS should 'fly' now!!!

(We'll get there boys and girls. SOON ENOUGH)!!!

[kaalilaatikko]

An excellent and elegant "engineer's" solution! I.e. it follows the original formula closely enough and is much more easier to calculate than what I tried to suggest. Just being curious, I looked at EURUSD with the original formula but used L = 0.02, taking the value by a quick look at the highest daily changes in the recent months. The form of this trend is of course exactly the same as with K = 3, but the scaling was quite close to that not using K/L at all. Some peaks are admittedly a bit "smoothed", but I have no doubt this last solution would not work, and look forward to getting the deep understanding of SI in full.

J.

[dpaterso]

Good (Tuesday) morning all!!!

kaalilaatikko:

I'm assuming you meant to say that you have no doubt that this last solution WILL work???

Yes: I checked the 'plot' of the ASI to ensure that the 'integrity' of the system was still intact and the HSP's and LSP's and closes above or below them are in the identical placed as they would WITH the 'limit' included. The only thing that 'appears' to be different is that the Trailing Index SAR now has 'meaning' to it and works 'across the board' with no manipulation of price formats etc. being necessary.

I've started using it today (last night) so let's see where we go with this.

Oh and by the way: thank you to those that gave input on this subject i.e. without those little 'tidbits of thought and ideas' I may not have been able to sort this out!!! Thank you.

[kaalilaatikko]

Originally Posted by dpaterso

I'm assuming you meant to say that you have no doubt that this last solution WILL work???

Yes, sure, the double negation just was too tough for me

[kaalilaatikko]

This is the 4th time I've scanned through this thread. 1st for VS, 2nd for ADX/ADXR, 3rd for TBPS and now for RTS and SIS. And this has been perhaps the most rewarding round. Now the posts I've earlier skipped make better sense. It's also a bit embarrassing to notice that I've brought up some issues that have already been discussed in the thread. I collected some remarks that have either been left unanswered or then I just wanted to have an opinion on the matter.

Dale 02-25-2008, 03:37 PM Post #50

If you're using the CSI to evaluate the different instruments and let's say that you find 10 instruments that are high on the CSI scale and you are being given entry signals. What then happens if a couple of days later those same 10 instruments are NO LONGER high on the CSI scale. Is THAT a signal to 'get out' OR do you just continue to stop and reverse as per the 'Trailing Index SAR' which (I think) is dangerous for the simple reason that should the CSI rating drop the instruments is then in effect becoming 'rangebound' so to just keep stopping and reversing if the instrument is now trading in a range is just looking to 'feed your broker' as they say in the 'classics'. Agree with me on this one? Ideas? Thoughts? Solutions? Try as I might I don't seem to come up with answers to the above that satisfy my logic so I'm 'throwing this open' to debate (which, after all, is the reason for this thread in the first place).

There have been some later posts about this as well. RTS is the first system right now for which I haven't done bactesting for most of the pairs, just looking at a short time period for some of the pairs that are the best ranging ones on the ADXR scale. I believe that this is the set where the system works best, and new pairs move into the set and old pairs move out of it, and you must be prepared to change pairs when needed. So at least I am in favour of doing this. For TBPS and DMS I selected the sets based on my backtesting. I then enter a trade only if a pair in my set has an adequate ADXR and is ranked high among the other pairs. For VS I don't use ADXR as has been adviced.

Dale 04-04-2008, 12:02 PM Post #172

On page 97: 'Fig. 806' and 'Fig. 807':

He shows the labels 'Significant High Swing Point' and 'Significant Low Swing Point' on those diagrams. What I don't 'get' is WHY are THOSE points 'Significant' and the other 'little swings' on those diagrams NOT
'Significant'???

Dale 04-07-2008, 11:59 AM #188

Referring to my post #172 (wherein I noted that I could not figure out WHY some points were 'Significant High Swing Points' or 'Significant Low Swing Points' while others were not designated as 'Significant'):

Well I THINK I've 'fathomed it out'. Right at the beginning of the book 'the man' says that a 'SIGNIFICANT HIGH POINT' is 'the highest price reached while in a Long trade' while a 'SIGNIFICANT LOW POINT' is 'the lowest price reached while in a Short trade' SO I've taken this to mean (as it relates to the SI System) that a 'SIGNIFICANT HSP' is the highest ASI value reached while in a Long trade and a 'SIGNIFICANT LSP' is the lowest ASI value reached while in a Short trade. Make sense???

chirules54 03-23-2008, 01:19 PM Post #121

1. When you enter initially (after above/below a significant swing point) the INITIAL SAR is the MOST RECENT
swing point in the other direction, so if you are going short then the SAR is the most recent high swing point,
correct?

2. After this, the SAR doesn't change until a new low swing point is made, which then makes the most recent high
swing point the new SAR. This repeats itself every time a new low swing point is made, correct?

Dale: As far as the SI System goes you've 'got it down 1000%'.

Am I getting this now completely correct? I'm not sure if the "new low swing point" is just an ordinary "low swing point" or should it actually be the "new significant low swing point?" Wilder uses the former term on p.
98, first paragraph, as well, and tells us: "Then keep the index sar at this point [the 1st lsp after a new hsp] until the ASI makes a new high." The following example and fig. 8.8 show how the index sar is set to point F as
told, but then it stays there even if there are LSPs G and H and HSPs in between. Only K changes the index sar, and that's because J is a significant HSP.

So, is the following wording of chirules54's question #2 correct: "After this, the SAR doesn't change until a new significant low swing point is made, which then makes the first high swing point after it as the new SAR. This repeats itself every time a new significant low swing point is made, correct?"

Dale 04-04-2008, 10:55 AM Post #170

I think that the Reaction Trend System is probably the most UNDERSTATED system in the book. The only thing that concerns me about the system is that 'B', 'O', and 'S' 'day labelling' i.e. you'll see in the book that it's based on the premise that we get three days up and two days down and I'm not ENTIRELY sure that this still applies. I promised myself that I would write some sort of indicator or calculator or something that will go through the data for a particular instrument and work out the 'average ratio' of up days to down days but I have not got around to it yet.

Wilder developed his systems originally for commodities where there always is just one sort to trade at a time. But in forex you have two in a pair, which made me to think this. The order of the currencies in a pair is fixed, but that order is just based on some agreement. Now if the order would have been opposite, the charts would have been vertically flipped as well. Let's assume that the 3 up/2 down rule works for a pair. Let's then assume that the International Forex Pair Naming Association decides to change the order of the currencies in that pair. This would mean that now the rule would read 3 down/2 up. Well, the example is ridiculous, but I'm trying to prove that maybe there is no reasoning why the 3/2 rule would work in forex as it might work for commodities. And therefore I'm questioning the basis for the BOSBOSBOS sequence. Why couldn't we use SOBSOBSOB as well? Or would BOSOBOSOBOSO be the best one? I have no further proof to take this further, but what do you, folks, think about
this?

And then...

Dale 05-19-2008, 08:25 AM Post # 410

I MAY have found a more 'scientific' method for the RTS 'BOS' sequencing:
Using Linear Channel Regression:
Method 1: ... [etc.]

Is this some sort of telepathy in the past tense or what? Complemented my pondering in such a brilliant way that I had to print it out right ahead for further examination.

balaji3003 05-03-2008, 03:31 PM #311

Just one question. Since "THE BOOK" has trading system that are completely technical, it is possible to completely automate the trading. ...right? Have you developed programs to fully automate trading strategies in the book?

Dale: As far as 'automation' is concerned: you are quite correct AND what's more I believe they would be EVEN MORE profitable if 'automated' ...

Personally I don't believe in a fully automated system where the computer would even open and close the positions. But I'm trying to develop myself a tool which I can configure to give me relevant signals on the pairs
I'm interested, and then it is my job to do the final evaluation on the signals and open and close the positions. I've read with interest about the posts where people have commented how much time they use on the systems daily. I hope to reach a point where the number of systems or pairs is not a limit and I could do my part in a quite short time. I've just made another idea that I haven't implemented yet: I am now able to reconstruct the daily candles so that the day closes on the hour I want. Now I could divide the pairs and systems in e.g. two and do the first half of the job at 6 AM and the second half at 7 AM, working all the time consistently on legitime daily candles. I believe that I should be able to handle quite a large amount of signals with a relatively little amount of effort, once I learn the internals of all of the systems well enough.

Dale 05-11-2008, 10:42 AM #374

I propose to add two new levels to the RT System i.e. S2 and B2.

Dale 05-17-2008, 11:34 AM #400

The new levels I am calling HSTOP and LSTOP and here are their respective formulae:

HSTOP = ( 3 * ( ( H + L + C ) / 3 ) ) - ( 3 * L ) + H
LSTOP = ( 3 * ( ( H + L + C ) / 3 ) ) - ( 3 * H ) + L

Sounds like yet another excellent idea! RTS seems to have started working for me, and I need to try this idea as well. Doing a little mathematics you can simplify the latter ones to:

HSTOP = ( H + L + C ) - ( 3 * L ) + H
LSTOP = ( H + L + C ) - ( 3 * H ) + L

Edit: maybe it's more elegant to beautify this a bit more:
HSTOP = ( 2 * H ) - ( 2 * L ) + C
LSTOP = ( 2 * L ) - ( 2 * H ) + C

A bit long post, but luckily I'm on vacation right now, more time to study these systems... there sure are no shortcuts for mastering this all.

J.

[dpaterso]

Good (Wednesday) morning!!!

kaalilaatikko:

WOW!!!

It's bee a LONG time since I've had to actually print out a post before answering it!!! Good on you!!!

I'm going to answer you this way:

Where I'm in agreement with you I'm not going to comment. I'm just going to comment on things where yours and my understanding or opinion differ otherwise I'll just be repeating what's already on the thread and what you've already posted.

CSI:

Remember that using the CSI to choose which pairs to trade (I'm going to post my responses ONLY as they pertain to forex pairs here) is 'first prize' BUT extremely difficult to set up and maintain on a daily basis i.e. you've constantly got to compensate for exchange rate differentials etc. etc. etc. and because you're working with up to four decimal places this makes it even more difficult. As I said: using the CSI is 'first prize' BUT as Wilder ALSO says: using ADXR is pretty much the same as using the CSI. The only difference is that the CSI is giving you and indication of the pairs with 'the best bang for buck' AND with good strong directional movement whereas ADXR on its own is giving you and indication of good strong directional movement ONLY. I don't use the CSI for forex trading but only ADXR. Look: sometimes you also have to take into account the 'reward/effort' ratio. In other words: if I'm netting a MINIMUM of 11% per month (which is my WORST performance to date as you know) WITHOUT spending hour upon hour maintaining the CSI 'proper' then I'm happy. This does not mean I'm lazy to do it BUT there just are time constraints and that's all there is to it. EVEN I have to sleep SOMETIME!!

Also (no matter whether you're using the CSI or ADXR only): remember that the CSI / ADXR indicates either which pairs will give you 'the best bang for buck with good strong directional movement' or just 'good strong directional movement' and it is these 'top rated' pairs that should be being traded at any one time. The ADX determines WHICH SYSTEM you should be trading FOR THOSE PAIRS!!! Don't get the two confused.

I would exit a PROFITABLE trade entered using a trend following system if ADX dropped below 20 -25 (or at WORST case if ADX dropped below +DI/-DI as per 'the book').

As you and I BOTH say: I'm not too concerned WHERE ADX is when entering a VS trade. It IS, however, interesting to note that when ADX is below 20 - 25 your VS trade tends to 'go nowhere' for a while (which although frustrating at least tells me that we're 'on track' with the ADX i.e. it's behaving EXACTLY as it should)!!!. I suppose that taking the ADX value into account when using the VS would make the difference between tying up margin for no good reason for long periods of time. It's a 'tradoff' though i.e. VS entry signals are very few and far between so you MAY get a VS entry signal when ADX is below 20 - 25 BUT by waiting for ADX to climb ABOVE 25 you could have missed the VS entry signal!!!

'Significant' points:

I'm no longer worried about whether a HSP or a LSP is 'significant' or not. If you look at the 'SIS red pages' you'll note the description of a HSP and a LSP and NOWHERE on the 'SIS red pages' is reference made to a 'significant' point. While my explanation or understanding of what the difference is between a HSP or a LSP or a 'significant' HSP or a 'significant' LSP MAY be correct JUST trading the HSP or LSP is working for me (and well now).

Index SAR and Trailing Index SAR:

This does indeed seem to be causing some confusion.

The INDEX SAR stays where it is UNTIL a new HSP or LSP has been formed. The TRAILING INDEX SAR 'tracks' the price on a daily basis.

For the purposes of explanation I've attached a chart of Fannie May because it's a 'textbook example' of WHY we need a Trailing Index SAR in the first place AND that it has to be CORRECTLY calculated.

If you take a nice look at the chart you'll see that WITHOUT the Trailing Index SAR (the pink line) which moves DOWN every day i.e. 'tracks' the price you'd only stop and reverse IF a new HSP was formed and that may not happen!!! On this chart: the INDEX SAR is indicated by the GREEN LINE at the moment.

RTS:

I no longer even attempt to use the 'BOS' sequencing. While other MAY disagree with me I'm convinced that's it's there and does exist. However: using the RTS with my additional levels and LRC's works fantastically well for me so again: if it's working THIS well WITHOUT the 'BOS' sequencing then why bother is the question. Remember also: 'pure' RTS will have to close at a loss sometimes whereas to date MY RTS/LRC 'system' has not had be realise a loss now for going on nine weeks.

Yes: there are other 'methods' that I've 'come up with' to try to make the 'BOS' sequencing 'easier' to find or more 'true' BUT I'm not convinced of the merits. For me: if price hits an RTS level OUTSIDE of the LRC I'll go long or short and then TP at either B1 or S1. Simple as that. Of course: you need to CONSTANTLY watch ADX!!! This is 'key' to this 'system'.

As far as the additional RTS levels: it, combined with the LRC's, is a 'winner' no question.

Automation:

I no longer believe that automating these systems will work AS well as maintaining them on a daily basis will. Nor do I any longer believe that they are to be traded 'blindly' and 'mechanically'. Over the past couple of months I've learned that a certain amount of 'common sense' is needed regardless of what system you're using. As an example: you'd not find me going short ANY of the USD/??? pairs right now for the simple reason that you can very easily and clearly see that there is strong resistance 'visible' on all of them. The only time I'd go short a USD/??? pair at present if IF the price closed BELOW those lines of resistance. Now this does NOT make me a 'technical guru' nor do I look at fundamentals BUT to me this is just 'common sense'.

As far as your 'manipulation' of the candles are concerned:

For a long time I had questions about 'which timezone presented the most important candles' and I was sure that the 'New York timezone close' was the most important to be looking at i.e. 00h00 New York time. Since those days though I'm of the opinion that the 'close' at the same time that the NYSE closes is the 'imporant' one and guess what: Delta falls 'smack bang' into this category!!! All I'm saying is this: I do understand your reasons for wanting to do this BUT I'm NOT convinced of the 'merit'.

One last thing:

O've been going through the TBPS again just looking at past trades. It's funny: normally a person tends to see 'only the good trades' but with the TBPS I find myself finding more BAD trades WITHOUT even looking for them. Personally: I don't think I can justify the 'risk/reward/effort' ration on that system. Also: I believe that it WILL work better on stocks and commodities because THEY DO move differently no question. For me: the VS/DMS/SIS/RTS are doing JUST FINE.

One last thing;

While I'm ALL FOR backtesting and analysing, just a word of advice: don't get TOO carried away with it i.e. just because (within reason) backtesting does not give you the desired results just remember that it's normally not indicative of future performance. What I'm saying is to NOT get 'fearful' of 'pulling the trigger' because your analsysis 'scares you' is all.

I hope that's covered everything!!!

[chdorry]

Reading Dale's post regarding the BOS sequencing led me to a thought as to why BOS *might not* apply to FX pairs. As BOS sequencing in commodities is specific to a SINGLE commodity. So for example if I take a long position on EUR/USD, I am buying EUR and selling USD. So I'm expecting the value of a EUR relative to the value of a USD to increase.

This can happen with an increase in the value of a EUR OR the decrease in the value of the USD. So as we are buying and selling on the same day (each individual currency) one could argue that for forex we'd have "dual BOS sequences" unless there was some way to identify if the position was expecting the EUR to increase (ie long the EUR) or the USD to decrease (short the USD). Since they are paired and the price is one relative to the other, I don't think there's any system that will differentiate which one is moving in your favor out of the pair.

What I mean is this:
Buying EUR gives this sequence:
B O S B O S B O S B O S

Selling USD gives this sequence:
S B O S B O S B O S B O

So looking at the above we are going to have days that are either B/S, O/B, or S/O depending on which currency in the pair you are actually looking at.

I'm going to have to look at this dual BOS sequence a little further this evening when I get home from work and see if this helps with the BOS sequencing applied to the RTS.

As of now I'm in agreement that while its there (but I'm not seeing it crystal clear) it doesn't really seem to have a great deal of impact on the profitability/tradeability of the RTS system.

Craig